US11003726B2 - Method, apparatus, and system for recommending real-time information - Google Patents

Method, apparatus, and system for recommending real-time information Download PDF

Info

Publication number
US11003726B2
US11003726B2 US15/868,729 US201815868729A US11003726B2 US 11003726 B2 US11003726 B2 US 11003726B2 US 201815868729 A US201815868729 A US 201815868729A US 11003726 B2 US11003726 B2 US 11003726B2
Authority
US
United States
Prior art keywords
interest
user
real
time information
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active, expires
Application number
US15/868,729
Other languages
English (en)
Other versions
US20180129749A1 (en
Inventor
Yucheng Hu
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Assigned to TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED reassignment TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: HU, YUCHENG
Publication of US20180129749A1 publication Critical patent/US20180129749A1/en
Application granted granted Critical
Publication of US11003726B2 publication Critical patent/US11003726B2/en
Active legal-status Critical Current
Adjusted expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2457—Query processing with adaptation to user needs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2457—Query processing with adaptation to user needs
    • G06F16/24575—Query processing with adaptation to user needs using context
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24—Querying
    • G06F16/245—Query processing
    • G06F16/2457—Query processing with adaptation to user needs
    • G06F16/24578—Query processing with adaptation to user needs using ranking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation

Definitions

  • the present disclosure relates to the field of communication technologies and, specifically, to a method, an apparatus, and a system for recommending real-time information.
  • Information recommendation refers to recommending, according to interests and behaviors of a user, information in which the user is interested.
  • Existing recommendation algorithms can be mainly classified into two classes: the first is a behavior-based recommendation algorithm, and the second is a content-based recommendation algorithm.
  • behavior-based recommendation algorithm in general, statistical analysis is performed on a user's behavior on information, similarities among information items in a recommendation pool are calculated, and then information items having a higher information similarity and corresponding to the user's behavior is recommended to the user.
  • content-based recommendation algorithm keywords are given for various information categories, and interests of a user are analyzed. Thus, a key word interested by the user can be determined and, based on the interested keyword and the keywords of the various information categories, a recommendation list is calculated to be recommended to the user.
  • the existing recommendation solutions often need to rely on participation of a large number of users, or do not consider any change in interests of users.
  • the timeliness is relatively poor.
  • the timeliness is extremely important. Therefore, for the real-time information, the recommendation effect of the existing recommendation solutions is not desirable.
  • Embodiments of the present invention provide a method, an apparatus, and a system for recommending real-time information, to improve timeliness, and flexibly and accurately recommend real-time information most-likely interested by a user currently to the user in time, thereby greatly improving the recommendation effect.
  • An embodiment of the present invention provides a method for recommending real-time information, including: obtaining user behavior data of a user; based on the user behavior data, respectively calculating a short-term interest, a long-term interest, and a real-time interest of the user; determining an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user; and recommending real-time information to the user based on the interest of the user.
  • the respectively calculating a short-term interest, a long-term interest, and a real-time interest of the user includes: based on the user behavior data, calculating a daily interest weight of the user for each day within a preset period to obtain daily interest weights, and attenuating the daily interest weights according to time to obtain the short-term interest of the user; based on the user behavior data, calculating an interest weight of the user within a preset time range to obtain the long-term interest of the user, wherein the preset time range is greater than one day; and based on the user behavior data, determining an interest weight of information currently clicked by the user to obtain the real-time interest of the user.
  • Another embodiment of the present invention provides a non-transitory computer-readable storage medium containing computer-executable instructions for, when executed by one or more processors, performing a method for recommending real-time information.
  • the method includes: obtaining user behavior data of a user; based on the user behavior data, respectively calculating a short-term interest, a long-term interest, and a real-time interest of the user; determining an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user; and recommending real-time information to the user based on the interest of the user.
  • the respectively calculating a short-term interest, a long-term interest, and a real-time interest of the user includes: based on the user behavior data, calculating a daily interest weight of the user for each day within a preset period to obtain daily interest weights, and attenuating the daily interest weights according to time to obtain the short-term interest of the user; based on the user behavior data, calculating an interest weight of the user within a preset time range to obtain the long-term interest of the user, wherein the preset time range is greater than one day; and based on the user behavior data, determining an interest weight of information currently clicked by the user to obtain the real-time interest of the user.
  • FIG. 1A is a schematic diagram of a system for recommending real-time information according to an embodiment of the present invention
  • FIG. 1B is a flowchart of a method for recommending real-time information according to an embodiment of the present invention
  • FIG. 2 is a flowchart of another method for recommending real-time information according to an embodiment of the present invention
  • FIG. 3A is a structural diagram of an apparatus for recommending real-time information according to an embodiment of the present invention.
  • FIG. 3B is a structural diagram of another apparatus for recommending real-time information according to an embodiment of the present invention.
  • FIG. 4 is a structural diagram of a computing system according to an embodiment of the present invention.
  • the embodiments of the present invention provide a method, an apparatus, and a system for recommending real-time information.
  • the system for recommending real-time information includes the apparatus for recommending real-time information provided in certain embodiments of the present invention.
  • the apparatus for recommending real-time information may be specifically integrated in a server, for example, a recommendation server.
  • the system for recommending real-time information may further include another device, for example, user equipment, a user server configured to store user behavior data, and an information server configured to store original real-time information, etc.
  • the recommendation server may obtain the user behavior data from the user server, respectively calculate a short-term interest, a long-term interest, and a real-time interest of a user according to the user behavior data, then determine an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user, obtain the real-time information from the information server, and recommend the real-time information, for example, news, to user equipment based on the interest of the user.
  • the short-term interest refers to an interest corresponding to an interest weight of the user that is calculated in a relatively short period. Specifically, an interest weight of the user of each day in a preset period (for example, 30 days) may be calculated according to the obtained user behavior data, so as to obtain a daily interest weight, and the daily interest weight is attenuated according to time to obtain the short-term interest.
  • the long-term interest refers to an interest corresponding to an interest weight of the user that is calculated in a relatively long period. For example, an interest weight of the user within one year may be calculated according to the user behavior data.
  • the real-time interest refers to an interest corresponding to a current interest weight of the user, for example, a keyword or a label currently clicked by the user.
  • an apparatus for recommending real-time information is provided.
  • the apparatus for recommending real-time information may be specifically integrated in a server, such as a recommendation server, etc.
  • the apparatus may perform a method for recommending real-time information which may include: obtaining user behavior data; respectively calculating a short-term interest, a long-term interest, and a real-time interest of a user according to the user behavior data; determining an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user; and recommending real-time information to the user based on the interest of the user.
  • the method for recommending real-time information may include the followings.
  • the user behavior data refers to related data that may be used for user behavior analysis, for example, browsing history, clicking history, and/or downloading history of the user.
  • the user behavior data may be stored in the apparatus for recommending real-time information, or may be stored in another device, for example, a user server.
  • the user behavior data may be obtained from the user server.
  • A according to the daily interest weights, determining an interest weight that currently needs to be attenuated.
  • a date difference between a date of the interest weight that needs to be attenuated and a current date may be determined; a product of the date difference and a preset attenuation coefficient is calculated, and a difference between 1 and the product is calculated; and then the interest weight that needs to be attenuated is multiplied by the difference to obtain the attenuated interest weight.
  • the attenuation coefficient may be set according to an actual application requirement, and details are not described herein.
  • C repeating A and B until all interest weights that need to be attenuated in the daily interest weights are attenuated. That is, attenuation processing is performed according to time on the interest weight of each day to obtain attenuated daily interest weight for each day within the preset period of time.
  • Attenuated interest weights e.g., attenuated daily interest weight for each day within the preset period of time
  • the preset period may be set according to an actual application requirement, which for example, may be generally set to 7 days, 15 days, or 30 days.
  • the preset time range may be set according to an actual application requirement, and is at least greater than one day, which, for example, may be set to one quarter, one year, or two years.
  • statistical analysis may be performed on monthly behavior of the user within one year from a current date according to the user behavior data; a monthly interest weights of each month can be calculated according to the monthly behavior of the user; an average interest weight within one year is calculated according to each monthly interest weight; and statistical analysis is performed on the average interest weight to obtain the long-term interest of the user.
  • each monthly interest weight may also be attenuated according to time to obtain attenuated monthly interest weights. Then statistical analysis is performed on these attenuated interest weights to obtain the long-term interest of the user.
  • a message currently clicked by the user includes a keyword (or a label) “NBA”, it is determined that the current interest of the user is “NBA”. Therefore, an interest weight of the keyword “NBA” may be calculated. Thus, similarly, the real-time interest of the user may be obtained.
  • 103 Based on the short-term interest, the long-term interest, and the real-time interest of the user, determining the interest of the user or the predicted interest of the user.
  • the short-term interest, the long-term interest, and the real-time interest of the user may be merged according to a preset rule, so as to obtain the interest of the user.
  • individual weights may be respectively set for the short-term interest, the long-term interest, and the real-time interest and, based on these respective weights, the short-term interest, the long-term interest, and the real-time interest are merged together.
  • a function may be set for the relationships among the short-term interest, the long-term interest, and the real-time interest, and then the short-term interest, the long-term interest, and the real-time interest are merged together by using the function.
  • Other ways may also be used.
  • the real-time information may include news, etc.
  • corresponding real-time information may be recalled from an inverted index of real-time information to obtain candidate recommendation information items and, based on the candidate recommendation information items, the real-time information is recommended to the user. More specifically:
  • A Calculating a matching degree between each real-time information item in the candidate recommendation information items and the interest of the user to obtain an interest correlativeness of the real-time information.
  • C Determining a click-through rate of each real-time information item in the candidate recommendation information items, and calculating a click-through model factor (CM) according to the click-through rate.
  • CM click-through model factor
  • D Determining the recommendation information from the candidate recommendation information items according to the interest correlativeness, the timeliness, and the click-through model factors.
  • the real-time information items in the candidate recommendation information items may be scored according to the interest correlativeness, the timeliness, and the click-through model factors, and then real-time information items whose scores are higher than a preset threshold are determined as the recommendation information.
  • the preset threshold may be set according to an actual application requirement.
  • information quality of each real-time information item in the candidate recommendation information items may be determined.
  • a quality factor of a news item may be determined by text recognition.
  • a junk article or an advertising article may have a low quality-score. That is, before determining the recommendation information from the candidate recommendation information items according to the interest correlativeness, the timeliness, and the click-through model factors, the information quality of each real-time information item in the candidate recommendation information items may also be determined.
  • the recommendation information from the candidate recommendation information items may be determined according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality by, for example, scoring the real-time information items in the candidate recommendation information items according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality; and determining real-time information items whose score are higher than a preset threshold as the recommendation information.
  • the inverted index of the real-time information items may be obtained by collecting original real-time information items and collecting statistics on the original real-time information items. That is, before recalling corresponding real-time information items from an inverted index of real-time information according to the interest of the user to obtain candidate recommendation information items, the followings may be performed, including: obtaining the original real-time information items from an original real-time information database; extracting features of the obtained original real-time information items; performing category prediction and topic prediction on the original real-time information items according to the extracted features to determine a category and a subject of each original real-time information; performing text field weighting after performing property weighting processing on content of the obtained original real-time information items to determine a keyword to which the original real-time information item belongs; and calculating an inverted index of original real-time information item in the original real-time information database according to the category, the subject, and the keyword of each original real-time information item to obtain the inverted index of the real-time information.
  • the followings may be specifically performed: determining, according to the interest of the user, a category, a subject, and/or a keyword in which the user is interested, and obtaining, from the inverted index of the real-time information, one or more original real-time information items that are the same as, similar to, or close to the category, the subject, and/or the keyword in which the user is interested, so as to obtain the candidate recommendation information items.
  • synonyms and/or near-synonyms may be set for words involved in the category, the subject, and/or the keyword in which the user is interested. If the category, the subject, and/or the keyword of the original real-time information item include words the same as these synonyms and/or near-synonyms, it is determined that the original real-time information item is original real-time information that is similar to or close to the category, the subject, and/or the keyword in which the user is interested.
  • user behavior data is obtained, a short-term interest, a long-term interest, and a real-time interest of a user are respectively calculated according to the user behavior data, then an interest of the user is determined according to the short-term interest, the long-term interest, and the real-time interest of the user, and real-time information is recommended to the user based on the interest of the user.
  • the interest of the user is calculated, not only the long-term interest of the user is taken into consideration, but also the short-term interest and the real-time interest of the user are taken into consideration, so as to reflect changes in interests of the user with time. Therefore, compared with the existing technology, real-time information in which the user is most-likely interested currently can be more flexibly and accurately recommended to the user in time, thereby greatly improving the recommendation effect while improving timeliness.
  • an interest correlativeness of the real-time information may be considered, but also timeliness of the real-time information, a click-through rate of a keyword, information quality, and the like may also be considered. Therefore, closeness of a relationship between the recommendation information and the interest of the user can be more accurately described, thereby further improving recommendation quality and the recommendation effect.
  • a recommendation server incorporating the apparatus for recommending real-time information is provided, and a method for recommending the real-time information is also provided.
  • a specific procedure of a method for recommending real-time information may include the followings.
  • a recommendation server obtains user behavior data from a user server.
  • the user behavior data refers to related data that may be used for user behavior analysis, for example, data such as browsing history, clicking history, and/or downloading history of a user.
  • the recommendation server calculates daily interest weights of the user within 30 days according to the user behavior data, to obtain daily interest weights, and attenuates the daily interest weights according to time, to obtain a short-term interest of the user. For example, this step may be specifically as follows:
  • A Determining, according to the daily interest weights, an interest weight that currently needs to be attenuated.
  • a date difference between a date of the interest weight that needs to be attenuated and a current date may be determined; a product of the date difference and a preset attenuation coefficient is calculated, and a difference between 1 and the product is calculated; and then the interest weight that needs to be attenuated is multiplied by the difference to obtain the attenuated interest weight.
  • final_weight k interesr_weight k *(1 ⁇ D * ⁇ ) where ⁇ represents the attenuation coefficient, D represents the date difference between the date of the interest weight that needs to be attenuated and the current date, interest_weight k represents the interest weight that needs to be attenuated, and final_weight k represents the attenuated interest weight; and the attenuation coefficient may be set according to an actual application requirement.
  • the current date is August 10
  • the recommendation server collects statistics on an interest weight of the user within one year from a current date according to the user behavior data to obtain a long-term interest of the user.
  • statistics may be collected on monthly behaviors of the user within one year from a current date according to the user behavior data; each monthly interest weight are calculated according to the monthly behaviors of the user; an average interest weight within one year is calculated according to each monthly interest weight; and statistics is collected on the average interest weight to obtain the long-term interest of the user.
  • Each monthly interest weight may be attenuated according to time to obtain attenuated interest weights. Then, statistics are collected on these attenuated interest weights to obtain the long-term interest of the user.
  • the recommendation server determines, according to the user behavior data, an interest weight of information currently clicked by the user to obtain a real-time interest of the user.
  • a weight of the keyword “NBA” may be calculated. The rest can be deduced by analogy, and the real-time interest of the user may be obtained.
  • the recommendation server determines an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user.
  • the short-term interest, the long-term interest, and the real-time interest of the user may be merged according to a preset rule to obtain the interest of the user.
  • individual weights may be respectively set for the short-term interest, the long-term interest, and the real-time interest and, based on these respective weights, the short-term interest, the long-term interest, and the real-time interest are merged together.
  • a function may be set for the relationships among the short-term interest, the long-term interest, and the real-time interest, and then the short-term interest, the long-term interest, and the real-time interest are merged together by using the function.
  • Other ways may also be used.
  • the recommendation server obtains an inverted index of news. For example, followings may be specifically performed: obtaining original news from an original real-time information database; extracting certain features of the obtained original news; performing category prediction and topic prediction on the original news according to the extracted features to determine a category and a subject of the original news; performing text field weighting after performing property weighting processing on content of the obtained original news to determine a keyword to which the original news belongs; and calculating an inverted index of original news in the original news base according to the category, the subject, and the keyword of the original news to obtain the inverted index of news.
  • the original real-time information database may be stored in the recommendation server, or may be stored in another device, for example, an information server.
  • the category of the original news may be predicted (i.e., category prediction) by using liblinear (a technology used for generating a categorizer) to obtain the category of the original news.
  • the subject of the original news may be predicted (that is, topic prediction) by using a subject model (e.g., LDA or latent Dirichlet allocation), that is, latent subject information in a document of the original news may be recognized by using LDA, to obtain the subject of the original news.
  • a subject model e.g., LDA or latent Dirichlet allocation
  • Property weighting processing may be performed on the content of the obtained original news by using term frequency-inverse document frequency (TF-IDF).
  • TF-IDF is a common weighting technology used for information retrieval and data mining.
  • TF-IDF is also a statistics method, used for evaluating an importance degree of a word for a document in a document set or a corpus.
  • the recommendation server recalls corresponding news from the inverted index of the news according to the interest of the user to obtain candidate recommendation information items.
  • a category, a subject, and/or a keyword in which the user is interested may be determined according to the interest of the user, and original news that is the same as, similar to, or close to the category, the subject, and/or the keyword in which the user is interested is obtained from the inverted index of the news to obtain the candidate recommendation information items.
  • the recommendation server recommends news to the user based on the candidate recommendation information items. For example, followings may be specifically performed.
  • A Calculating a matching degree between each news item in the candidate recommendation information items and the interest of the user to obtain an interest correlativeness between the news and the user.
  • C Determining a click-through rate of each news item in the candidate recommendation information items, and calculating a click-through model factor (CM) according to the click-through rate.
  • CM click-through model factor
  • a quality factor of a news item may be determined by text recognition.
  • a junk article or an advertising article has a low quality-score.
  • Determining the recommendation news from the candidate recommendation information items according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality For example, the followings may be specifically performed: scoring the news in the candidate recommendation information items according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality; and determining news whose score is higher than a preset threshold as the recommendation news.
  • user behavior data is obtained, a short-term interest, a long-term interest, and a real-time interest of a user are respectively calculated according to the user behavior data, then an interest of the user is determined according to the short-term interest, the long-term interest, and the real-time interest of the user, and news is recommended to the user based on the interest of the user.
  • the interest of the user is calculated, not only the long-term interest of the user is taken into consideration, but also the short-term interest and the real-time interest of the user are taken into consideration, to reflect changes in interests of the user with time. Therefore, compared with the existing technology, news in which the user is most-likely interested currently can be more flexibly and accurately recommended to the user in time, thereby greatly improving a recommendation effect while improving timeliness.
  • an apparatus for recommending real-time information includes: an obtaining unit 301 , a calculation unit 302 , a determining unit 303 , and a recommendation unit 304 .
  • the obtaining unit 301 is configured to obtain user behavior data.
  • the user behavior data refers to related data that may be used for user behavior analysis, for example, browsing history, clicking history, and/or downloading history of a user.
  • the calculation unit 302 is configured to respectively calculate a short-term interest, a long-term interest, and a real-time interest of a user according to the user behavior data.
  • the calculation unit 302 may include: a first calculation subunit, a second calculation subunit, and a third calculation subunit.
  • the first calculation subunit is configured to: calculate a daily interest weight of the user in a preset period according to the user behavior data to obtain a daily interest weight, and to attenuate the daily interest weight according to time to obtain the short-term interest of the user.
  • the first calculation subunit may be specifically configured to: determine, according to the obtained daily interest weight, an interest weight that currently needs to be attenuated; attenuate, according to time, the interest weight that needs to be attenuated to obtain an attenuated interest weight; continue determining and attenuating the interest weight that currently needs to be attenuated until attenuation of all interest weights that need to be attenuated in the daily interest weight is finished; and collect statistics on all obtained attenuated interest weights to obtain the short-term interest of the user.
  • the first calculation subunit may be specifically configured to: determine a date difference between a date of the interest weight that needs to be attenuated and a current date; calculate a product of the date difference and a preset attenuation coefficient, and calculate a difference between 1 and the product; and multiply the interest weight that needs to be attenuated by the difference to obtain the attenuated interest weight. More details may be referred to the foregoing method embodiment.
  • the preset period may be set according to an actual application requirement, for example, may be generally set to 7 days, 15 days, or 30 days.
  • the second calculation subunit is configured to calculate an interest weight of the user within a preset time range according to the user behavior data to obtain the long-term interest of the user.
  • the preset time range may be set according to an actual application requirement, and is at least greater than one day, for example, may be set to one quarter, one year, or two years.
  • the second calculation subunit may be specifically configured to: collect statistics on monthly behaviors of the user within one year from a current date according to the user behavior data; calculate each monthly interest weight according to the monthly behaviors of the user; calculate an average interest weight within one year according to each monthly interest weight; and collect statistics on the average interest weight, to obtain the long-term interest of the user.
  • the second calculation subunit may also attenuate each monthly interest weight, according to time, to obtain attenuated interest weights; and then collect statistics on these attenuated interest weights to obtain the long-term interest of the user.
  • the third calculation subunit is configured to determine, according to the user behavior data, an interest weight of information currently clicked by the user to obtain the real-time interest of the user.
  • the determining unit 303 is configured to determine an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user.
  • the determining unit 303 may be specifically configured to merge the short-term interest, the long-term interest, and the real-time interest of the user according to a preset rule, to obtain the interest of the user.
  • individual weights may be respectively set for the short-term interest, the long-term interest, and the real-time interest and, based on these respective weights, the short-term interest, the long-term interest, and the real-time interest are merged together.
  • a function may be set for the relationships among the short-term interest, the long-term interest, and the real-time interest, and then the short-term interest, the long-term interest, and the real-time interest are merged together by using the function.
  • Other ways may also be used.
  • the recommendation unit 304 is configured to recommend real-time information to the user based on the interest of the user.
  • the real-time information may be specifically information such as news.
  • the recommendation unit 304 may include a recall subunit and a recommendation subunit.
  • the recall subunit is configured to recall corresponding real-time information from an inverted index of real-time information according to the interest of the user to obtain candidate recommendation information items.
  • the recommendation subunit is configured to recommend the real-time information to the user based on the candidate recommendation information items.
  • the recommendation subunit may be configured to: calculate a matching degree between each real-time information item in the candidate recommendation information items and the interest of the user to obtain an interest correlativeness of the real-time information; determine freshness of each real-time information item by determining a release time of each real-time information item in the candidate recommendation information items to obtain timeliness of the real-time information; determine a click-through rate of each real-time information item in the candidate recommendation information items, and calculate a click-through model factor according to the click-through rate; determine recommendation information from the candidate recommendation information items according to the interest correlativeness, the timeliness, and the click-through model factors; and recommend the recommendation information to the user.
  • information quality of each real-time information item in the candidate recommendation information items may be determined.
  • a quality factor of a news item may be determined by text recognition.
  • a junk article or an advertising article has a low quality-score.
  • the recommendation subunit may be specifically configured to: calculate a matching degree between each real-time information item in the candidate recommendation information items and the interest of the user, to obtain an interest correlativeness of the real-time information; determine freshness of each real-time information item by determining a release time of each real-time information item in the candidate recommendation information items to obtain timeliness of the real-time information; determine a click-through rate of each real-time information item in the candidate recommendation information items, and calculate a click-through model factor according to the click-through rate; determine information quality of each real-time information item in the candidate recommendation information items; determine recommendation information from the candidate recommendation information items according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality; and recommend the recommendation information to the user.
  • the inverted index of the real-time information may be obtained by collecting original real-time information and collecting statistics on the original real-time information. That is, as shown in FIG. 3B , the apparatus for recommending real-time information may further include an inverted index determining unit 305 .
  • the inverted index determining unit 305 may be configured to: obtain original real-time information from an original real-time information database; extract a feature of the obtained original real-time information; perform category prediction and topic prediction on the original real-time information according to the extracted feature, to determine a category and a subject of the original real-time information; perform text field weighting after performing property weighting processing on content of the obtained original real-time information, to determine a keyword to which the original real-time information belongs; and calculate an inverted index of original real-time information in the original real-time information database according to the category, the subject, and the keyword of the original real-time information, to obtain the inverted index of the real-time information.
  • the recall subunit is specifically configured to: determine, according to the interest of the user, a category, a subject, and/or a keyword in which the user is interested, and obtain, from the inverted index of the real-time information, original real-time information that is the same as, similar to, or close to the category, the subject, and/or the keyword in which the user is interested to obtain the candidate recommendation information items.
  • the foregoing units may be implemented as independent entities, or may be arbitrarily combined, and implemented as a same entity or several entities.
  • the foregoing units refer to the foregoing method embodiment, and details are not described herein again.
  • the apparatus for recommending real-time information may be specifically integrated into a server, for example, a recommendation server.
  • the obtaining unit 301 of the apparatus for recommending real-time information may obtain user behavior data, the calculation unit 302 respectively calculates a short-term interest, a long-term interest, and a real-time interest of a user according to the user behavior data, then the determining unit 303 determines an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user, and the recommendation unit 304 recommends real-time information to the user based on the interest of the user.
  • the interest of the user is calculated, not only the long-term interest of the user is taken into consideration, but also the short-term interest and the real-time interest of the user are taken into consideration, to reflect changes in interests of the user with time. Therefore, compared with the existing technology, real-time information in which the user is most-likely interested currently can be more flexibly and accurately recommended to the user in time, thereby greatly improving the recommendation effect while improving timeliness.
  • the recommendation unit 304 calculates the candidate recommendation information items, not only an interest correlativeness of the real-time information may be considered, but also timeliness of the real-time information, a click-through rate of a keyword, information quality, and the like may also be considered. Therefore, closeness of a relationship between the information and the interest of the user can be more accurately described, thereby further improving recommendation quality and the recommendation effect.
  • this embodiment of the present invention further provides a system for recommending real-time information, which may include the apparatus for recommending real-time information provided in any one of the embodiments of the present invention.
  • the apparatus for recommending real-time information may be specifically integrated into a server, for example, a recommendation server.
  • a server for example, a recommendation server.
  • the following description is provided using the recommendation server as an example.
  • the recommendation server is configured to: obtain user behavior data; respectively calculate a short-term interest, a long-term interest, and a real-time interest of a user according to the user behavior data; determine an interest of the user according to the short-term interest, the long-term interest, and the real-time interest of the user; and recommend real-time information to the user based on the interest of the user.
  • a daily interest weight of the user in a preset period is calculated according to the user behavior data, to obtain a daily interest weight, and the daily interest weight is attenuated, according to time, to obtain the short-term interest of the user.
  • the preset period may be set according to an actual application requirement, for example, may be generally set to 7 days, 15 days, or 30 days.
  • an interest weight of the user within a preset time range is calculated according to the user behavior data to obtain the long-term interest of the user.
  • the preset time range may be set according to an actual application requirement, and is at least greater than one day, for example, may be set to one quarter, one year, or two years.
  • an interest weight of information currently clicked by the user is determined according to the user behavior data to obtain the real-time interest of the user.
  • the recommendation server may specifically recall corresponding real-time information from an inverted index of real-time information according to the interest of the user, so as to obtain candidate recommendation information items, and then recommend the real-time information to the user based on the candidate recommendation information items.
  • the recommendation server may calculate parameters such as interest correlativeness, timeliness, click-through model factors, and information quality of the real-time information, and then determine recommendation information from the candidate recommendation information items according to the interest correlativeness, the timeliness, the click-through model factors, and the information quality.
  • the inverted index of the real-time information may be obtained by collecting original real-time information and collecting statistics on the original real-time information.
  • the recommendation server may be configured to obtain the original real-time information from an original real-time information database; extract a feature of the obtained original real-time information; perform category prediction and topic prediction on the original real-time information according to the extracted feature, to determine a category and a subject of the original real-time information; perform text field weighting after performing property weighting processing on content of the obtained original real-time information, to determine a keyword to which the original real-time information belongs; and calculate an inverted index of original real-time information in the original real-time information database according to the category, the subject, and the keyword of the original real-time information, to obtain the inverted index of the real-time information.
  • system for recommending real-time information may further include another device, for example, user equipment, and optionally, may further include a user server and an information server.
  • the user equipment may be configured to receive the real-time information recommended by the recommendation server.
  • the user server may be configured to: store the user behavior data, and provide the user behavior data for the recommendation server.
  • the information server may be configured to: store the original real-time information, and provide the original real-time information for the recommendation server.
  • the system for recommending real-time information may include the apparatus for recommending real-time information provided in any one of the embodiments of the present invention and, therefore, can achieve the beneficial effects that can be achieved by the apparatus for recommending real-time information provided in any one of the embodiments of the present invention.
  • the program may be stored in a computer readable storage medium.
  • the storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk, an optical disc, or the like.
  • FIG. 4 illustrates an exemplary computing system for implementing the various modules, units, apparatus, and systems.
  • the computing system 400 includes: a display 401 , a processor 402 , a memory 403 , an input device 404 (for example, a peripheral device such as a collection device including a camera, a microphone, and a headset; a mouse, a joystick, or a desktop computer keyboard; or a physical keyboard or a touchscreen on a notebook computer or a tablet computer), an output device 405 (for example, an audio output device or a video output device including a speaker, a headset, and the like), a bus 406 , and a networking device 407 .
  • a peripheral device such as a collection device including a camera, a microphone, and a headset
  • a mouse, a joystick, or a desktop computer keyboard or a physical keyboard or a touchscreen on a notebook computer or a tablet computer
  • an output device 405 for example, an audio output device or a video output device including a
  • the processor 402 may include any appropriate hardware processing unit, such as a central processing unit (CPU), a graphic processing unit (GPU), or a microcontroller, etc.
  • the processor 402 , the memory 403 , the input device 404 , the display 401 , and the networking device 407 are connected by using the bus 406 , and the bus 406 is used for data transmission and communication between the processor 402 , the memory 403 , the display 401 , and the networking device 407 .
  • the input device 404 is mainly configured to obtain an input operation of a user, and the input device 404 may include any appropriate device, such as a mouse, a keyboard, or a touchscreen, etc.
  • the networking device 407 is used to connect to other devices and systems.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
US15/868,729 2015-09-08 2018-01-11 Method, apparatus, and system for recommending real-time information Active 2037-05-20 US11003726B2 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
CN201510564481.3A CN106503014B (zh) 2015-09-08 2015-09-08 一种实时信息的推荐方法、装置和系统
CN201510564481.3 2015-09-08
PCT/CN2016/078482 WO2017041484A1 (zh) 2015-09-08 2016-04-05 一种实时信息的推荐方法、装置和系统

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/078482 Continuation WO2017041484A1 (zh) 2015-09-08 2016-04-05 一种实时信息的推荐方法、装置和系统

Publications (2)

Publication Number Publication Date
US20180129749A1 US20180129749A1 (en) 2018-05-10
US11003726B2 true US11003726B2 (en) 2021-05-11

Family

ID=58239117

Family Applications (1)

Application Number Title Priority Date Filing Date
US15/868,729 Active 2037-05-20 US11003726B2 (en) 2015-09-08 2018-01-11 Method, apparatus, and system for recommending real-time information

Country Status (3)

Country Link
US (1) US11003726B2 (zh)
CN (1) CN106503014B (zh)
WO (1) WO2017041484A1 (zh)

Families Citing this family (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108810056B (zh) * 2017-05-04 2021-05-14 腾讯科技（北京）有限公司 信息推送方法及装置
CN107545065B (zh) * 2017-09-08 2021-08-13 北京奇艺世纪科技有限公司 一种用户兴趣校准方法、系统及个性化推荐系统
CN110020117B (zh) * 2017-09-29 2022-05-03 北京搜狗科技发展有限公司 一种兴趣信息获取方法、装置及电子设备
CN107844525A (zh) * 2017-10-12 2018-03-27 广州艾媒数聚信息咨询股份有限公司 一种基于用户行为的资讯个性化推荐方法、系统及装置
CN107844598A (zh) * 2017-11-22 2018-03-27 广州优视网络科技有限公司 内容推荐方法、装置及计算机设备
CN110431535A (zh) * 2018-01-22 2019-11-08 华为技术有限公司 一种用户画像的生成方法及装置
CN108537567B (zh) * 2018-03-06 2020-08-07 阿里巴巴集团控股有限公司 一种目标用户群体的确定方法和装置
CN108805622B (zh) * 2018-06-11 2021-11-09 深圳乐信软件技术有限公司 商品推荐方法、装置、设备及存储介质
WO2020000207A1 (zh) * 2018-06-26 2020-01-02 深圳市爱的网络科技有限公司 用户兴趣采集方法、装置、计算机装置及计算机可读存储介质
CN109189951B (zh) * 2018-07-03 2021-09-03 南京尚网网络科技有限公司 一种多媒体资源推荐方法、设备及存储介质
CN109460514B (zh) * 2018-11-02 2024-06-14 北京京东尚科信息技术有限公司 用于推送信息的方法和装置
CN111241381A (zh) * 2018-11-28 2020-06-05 北京奇虎科技有限公司 信息推荐方法、装置、电子设备及计算机可读存储介质
CN109993570B (zh) * 2019-01-14 2023-09-01 深圳市东信时代信息技术有限公司 一种定向投放移动广告的方法及系统
CN110275952A (zh) * 2019-05-08 2019-09-24 平安科技（深圳）有限公司 基于用户短期兴趣的新闻推荐方法、装置及介质
CN110489639B (zh) * 2019-07-15 2022-10-25 北京奇艺世纪科技有限公司 一种内容推荐方法及装置
CN110516159B (zh) * 2019-08-30 2022-12-20 北京字节跳动网络技术有限公司 一种信息推荐方法、装置、电子设备及存储介质
CN110765309B (zh) * 2019-10-09 2023-09-01 上海麦克风文化传媒有限公司 一种基于参数配置的推荐系统召回方法及系统
CN110765310B (zh) * 2019-10-09 2024-01-30 上海麦克风文化传媒有限公司 一种基于参数配置的音频专辑推荐方法及系统
US11263174B2 (en) 2019-11-08 2022-03-01 International Business Machines Corporation Reducing resource consumption in container image management
CN111061945B (zh) * 2019-11-11 2023-06-27 汉海信息技术(上海)有限公司 推荐方法、装置、电子设备，存储介质
CN111209474B (zh) * 2019-12-27 2023-04-28 广东德诚科教有限公司 在线课程的推荐方法、装置、计算机设备和存储介质
CN111310033B (zh) * 2020-01-23 2023-05-30 山西大学 基于用户兴趣漂移的推荐方法及推荐装置
CN111444419A (zh) * 2020-03-02 2020-07-24 平安国际智慧城市科技股份有限公司 资源推荐方法、装置、计算机设备和存储介质
CN111369324B (zh) * 2020-03-12 2024-01-23 苏州大学 一种目标信息确定方法、装置、设备及可读存储介质
CN111444428B (zh) * 2020-03-27 2022-08-30 腾讯科技（深圳）有限公司 基于人工智能的信息推荐方法、装置、电子设备及存储介质
CN112084404B (zh) * 2020-09-01 2024-03-01 北京百度网讯科技有限公司 内容推荐方法、装置、设备和介质
CN113744016B (zh) * 2020-11-04 2024-05-24 北京沃东天骏信息技术有限公司 一种对象推荐方法及装置、设备、存储介质
WO2022201435A1 (ja) * 2021-03-25 2022-09-29 日本電信電話株式会社 情報処理装置、推定方法およびプログラム
CN113486250B (zh) * 2021-07-28 2023-09-05 中移（杭州）信息技术有限公司 内容推荐方法、装置、设备及计算机可读存储介质
CN115048586B (zh) * 2022-08-11 2023-02-21 广东工业大学 一种融合多特征的新闻推荐方法及系统
CN116049535A (zh) * 2022-08-18 2023-05-02 荣耀终端有限公司 信息推荐方法、装置、终端装置及存储介质
CN117828193B (zh) * 2024-03-04 2024-05-17 山东省计算中心（国家超级计算济南中心） 基于多兴趣半联合学习兴趣推荐方法、系统、设备及介质

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070118498A1 (en) * 2005-11-22 2007-05-24 Nec Laboratories America, Inc. Methods and systems for utilizing content, dynamic patterns, and/or relational information for data analysis
CN102663627A (zh) 2012-04-26 2012-09-12 焦点科技股份有限公司 个性化推荐方法
CN102681999A (zh) 2011-03-08 2012-09-19 阿里巴巴集团控股有限公司 一种用户行为信息收集及信息发送方法及装置
CN104090990A (zh) 2014-07-31 2014-10-08 北京奇虎科技有限公司 新闻推送方法和系统
US20140365468A1 (en) * 2013-06-10 2014-12-11 Microsoft Corporation News Results through Query Expansion
CN104246751A (zh) 2011-12-02 2014-12-24 Kddi株式会社 推荐装置、推荐系统、推荐方法以及程序
US20150088911A1 (en) 2013-09-25 2015-03-26 Alibaba Group Holding Limited Method and system for extracting user behavior features to personalize recommendations
US8996530B2 (en) * 2012-04-27 2015-03-31 Yahoo! Inc. User modeling for personalized generalized content recommendations
US20160239738A1 (en) * 2013-10-23 2016-08-18 Tencent Technology (Shenzhen) Company Limited Question recommending method, apparatus and system
US9514133B1 (en) * 2013-06-25 2016-12-06 Jpmorgan Chase Bank, N.A. System and method for customized sentiment signal generation through machine learning based streaming text analytics

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8001105B2 (en) * 2006-06-09 2011-08-16 Ebay Inc. System and method for keyword extraction and contextual advertisement generation
US9715543B2 (en) * 2007-02-28 2017-07-25 Aol Inc. Personalization techniques using image clouds
JP4433326B2 (ja) * 2007-12-04 2010-03-17 ソニー株式会社 情報処理装置および方法、並びにプログラム

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070118498A1 (en) * 2005-11-22 2007-05-24 Nec Laboratories America, Inc. Methods and systems for utilizing content, dynamic patterns, and/or relational information for data analysis
CN102681999A (zh) 2011-03-08 2012-09-19 阿里巴巴集团控股有限公司 一种用户行为信息收集及信息发送方法及装置
CN104246751A (zh) 2011-12-02 2014-12-24 Kddi株式会社 推荐装置、推荐系统、推荐方法以及程序
CN102663627A (zh) 2012-04-26 2012-09-12 焦点科技股份有限公司 个性化推荐方法
US8996530B2 (en) * 2012-04-27 2015-03-31 Yahoo! Inc. User modeling for personalized generalized content recommendations
US20140365468A1 (en) * 2013-06-10 2014-12-11 Microsoft Corporation News Results through Query Expansion
US9514133B1 (en) * 2013-06-25 2016-12-06 Jpmorgan Chase Bank, N.A. System and method for customized sentiment signal generation through machine learning based streaming text analytics
US20150088911A1 (en) 2013-09-25 2015-03-26 Alibaba Group Holding Limited Method and system for extracting user behavior features to personalize recommendations
US20160239738A1 (en) * 2013-10-23 2016-08-18 Tencent Technology (Shenzhen) Company Limited Question recommending method, apparatus and system
CN104090990A (zh) 2014-07-31 2014-10-08 北京奇虎科技有限公司 新闻推送方法和系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
The World Intellectual Property Organization (WIPO) International Search Report for PCT/CN2016/078482 dated Jul. 8, 2016 5 Pages (including translation).

Also Published As

Publication number Publication date
CN106503014B (zh) 2020-08-07
US20180129749A1 (en) 2018-05-10
WO2017041484A1 (zh) 2017-03-16
CN106503014A (zh) 2017-03-15

Similar Documents

Publication Publication Date Title
US11003726B2 (en) Method, apparatus, and system for recommending real-time information
US20230359680A1 (en) Personalized search filter and notification system
US20190197416A1 (en) Information recommendation method, apparatus, and server based on user data in an online forum
CN108170692B (zh) 一种热点事件信息处理方法和装置
US9830386B2 (en) Determining trending topics in social media
US7860878B2 (en) Prioritizing media assets for publication
US10891322B2 (en) Automatic conversation creator for news
US10248715B2 (en) Media content recommendation method and apparatus
Shi et al. Learning-to-rank for real-time high-precision hashtag recommendation for streaming news
US10482142B2 (en) Information processing device, information processing method, and program
US20180089325A1 (en) Method, Apparatus and Client of Processing Information Recommendation
US11640420B2 (en) System and method for automatic summarization of content with event based analysis
KR20140119269A (ko) 소셜 미디어 분석을 기반으로 복합이슈를 탐지하기 위한 장치, 시스템 및 그 방법
CN113688310A (zh) 一种内容推荐方法、装置、设备及存储介质
AU2017334864B2 (en) Systems and methods for providing a social media knowledge base
CN110750707A (zh) 关键词推荐方法、装置和电子设备
US20230112385A1 (en) Method of obtaining event information, electronic device, and storage medium
US10733221B2 (en) Scalable mining of trending insights from text
US11822609B2 (en) Prediction of future prominence attributes in data set
Lim et al. ClaimFinder: A Framework for Identifying Claims in Microblogs.
JP5844887B2 (ja) 通信ネットワークを通じたビデオ・コンテンツ検索のための支援
JP2016139216A (ja) 話題語ランキング装置、話題語ランキング方法、およびプログラム
US11907657B1 (en) Dynamically extracting n-grams for automated vocabulary updates
CN112818221B (zh) 实体的热度确定方法、装置、电子设备及存储介质
CN111723201A (zh) 一种用于文本数据聚类的方法和装置

Legal Events

Date Code Title Description
AS Assignment

Owner name: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, CHINA

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HU, YUCHENG;REEL/FRAME:044602/0618

Effective date: 20171227

Owner name: TENCENT TECHNOLOGY (SHENZHEN) COMPANY LIMITED, CHI

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HU, YUCHENG;REEL/FRAME:044602/0618

Effective date: 20171227

FEPP Fee payment procedure

Free format text: ENTITY STATUS SET TO UNDISCOUNTED (ORIGINAL EVENT CODE: BIG.); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY

STPP Information on status: patent application and granting procedure in general

Free format text: DOCKETED NEW CASE - READY FOR EXAMINATION

STPP Information on status: patent application and granting procedure in general

Free format text: NON FINAL ACTION MAILED

STPP Information on status: patent application and granting procedure in general

Free format text: NOTICE OF ALLOWANCE MAILED -- APPLICATION RECEIVED IN OFFICE OF PUBLICATIONS

STPP Information on status: patent application and granting procedure in general

Free format text: PUBLICATIONS -- ISSUE FEE PAYMENT VERIFIED

STCF Information on status: patent grant

Free format text: PATENTED CASE